Three answers compete for what to do about a health baseline that will not shift: measure harder, rest properly, or build the doing. Only one of the three has large trials standing behind it, and it is not the one with the subscription.
Cochrane has put general health checks through two full reviews and arrived at the same verdict twice. Krogsbøll and colleagues published the pooled result in the BMJ in 2012, drawing on the trials with usable outcome data and covering 182,880 adults. Deaths did not fall. Not deaths overall, not deaths from cardiovascular causes, not deaths from cancer. What did rise, in one of the included trials, was the count of new diagnoses each participant carried away. Seven years later the update reached the same place. On fatal and non-fatal ischaemic heart disease, pooling four trials and 164,881 people, the reviewers found little or no effect and graded their certainty as high. Their summary sentence was one line long and it has never been withdrawn: health checks of this general kind are unlikely to help.
Now set that against your own filing. Most people who read this have a folder somewhere. Blood panels stretching back years, a device on the wrist since about 2021, possibly an imaging report. The podcasts have been listened to and the vocabulary is fluent. And the folder has almost certainly grown faster than the person holding it.
Nobody should read that as a character flaw. It is what a particular model of health produces, reliably, and the model is worth naming before anyone spends another five years inside it.
This is an educational and strategic perspective, not personal medical advice. The views are the author's own and not statements by Atlas Cove Lda.
The promise camp one made on the trials' behalf
The first camp has a simple instruction: get more numbers. Underneath it sits an assumption about where the bottleneck lies, namely that people fail to act because they cannot see their own position clearly enough. Show them the reading and the behaviour follows. In most areas of a capable adult's life that assumption is sound, which is exactly why it survives here, where it mostly is not.
Take the strongest evidence this camp has, which is the wearable literature, and read it precisely. Ferguson and a team at the University of South Australia published an umbrella review in The Lancet Digital Health in 2022, identifying 39 systematic reviews and meta-analyses across roughly 164,000 participants. Trackers did shift behaviour. The authors' rounded summary was about 1,800 additional steps daily, something like 40 more minutes of walking, and a reduction of around a kilogram in body weight. Those are genuine effects and pretending otherwise would be dishonest.
Follow the same review one step further downstream, though. Effects on cholesterol, glycated haemoglobin and blood pressure came out typically small and frequently not statistically significant. Systolic blood pressure is a partial exception worth naming, since three of five meta-analyses did report a fall of roughly 2 to 4 mmHg. The pattern is otherwise consistent: the device shifted the thing it counts, and left largely untouched the physiology that was the reason for buying it.
Then there is a trial this camp rarely brings up. The IDEA study appeared in JAMA in 2016 and randomised its participants at the start, not at the end. Young adults with overweight or obesity, 471 assigned and 470 analysed, all received the same intensive diet and activity counselling through the first six months. From month seven the arms separated: one group added a wearable device with a web interface, the other carried on with web-based self-monitoring. At 24 months the wearable group had lost 3.5 kg and the self-monitoring group 5.9 kg, a difference of 2.4 kg in favour of not having the device. One trial in one population, and building a worldview on it would be foolish. It does, at the very least, retire the idea that bolting measurement onto a working intervention makes that intervention stronger.
Imaging deserves the same treatment. A systematic review by Gibson and colleagues in the BMJ in 2018 brought together 32 studies covering MRI of the brain and of the body, all in adults reporting nothing wrong. Potentially serious incidental findings turned up in 3.9% of scans, climbing to 12.8% when the count also took in findings whose seriousness was unclear, and about half of the serious category were suspected malignancies. Kwee and Kwee, reviewing whole-body screening MRI in the Journal of Magnetic Resonance Imaging in 2019, put the pooled false-positive proportion near 16%. Heterogeneity in both reviews is substantial and the confidence intervals are very wide, so these belong in a sentence as ranges rather than as settled figures. The direction survives the imprecision. Point a highly sensitive instrument at a population with nothing wrong and most of what it flags turns out to be nothing, and each flag costs a follow-up appointment, a waiting period, and several weeks of somebody's equanimity.
What camp one delivered to a generation of careful, well-resourced adults was a filing system, a recurring charge, and a background hum of worry. A different body was the one item not reliably included.
What sits in the gap, and it isn't facts
The mechanism here has been measured, repeatedly, and it explains the whole pattern.
Webb and Sheeran pulled together 47 experimental tests for Psychological Bulletin in 2006 and found that shifting someone's intention by a medium-to-large amount (d = 0.66) shifted their actual behaviour by a small-to-medium amount (d = 0.36). Broader reviews of health behaviour tend to land on intention explaining something like a fifth of the variance in what gets done. Move a person's intention a long way and their behaviour moves roughly half as far. Every product built on informing, persuading or alerting is casting into that shallower half.
Care is needed at this point, because the careless version of this argument says measurement is useless, and the evidence says nothing of the sort.
Harkin and colleagues, also in Psychological Bulletin, pooled 138 experimental studies with 19,951 participants between them in 2016. Prompting people to monitor progress towards a goal did promote reaching it, at d = 0.40, and the effect ran through the change in how often they monitored. Progress monitoring works, and it works for reasons that are legible.
The moderators carry the argument. Effects were larger where the outcome was physically written down, and larger again where it was reported to somebody or made public. Which describes a very particular activity: a goal at one end, a record in the middle, another person at the far end. It bears almost no resemblance to a device quietly accumulating readings that get glanced at on a commute.
Camp one borrowed the standing that progress monitoring had earned and then shipped something built to a different shape. Data collection with nothing fastened to the other end. No goal, no witness, no consequence attached to a poor reading. That version was never the thing the trials tested, which is why it has no evidence behind it and why the absence is unsurprising rather than mysterious.
The rest camp answers a question most people didn't ask
The counter-movement arrived on cue. Drawer the wrist device. Log off. Rest. Regulate the nervous system. The claim underneath is that recovery was the missing input all along.
For some readers that will be precisely correct, and this needs saying plainly, because exhaustion of that kind is real and nobody carrying it is helped by a founder recommending more effort.
What a restful break does over time, though, has been measured and the picture is not ambiguous. Speth, Wendsche and Wegge pooled 13 studies covering 1,428 employees for European Psychologist in 2023, with holidays averaging 11 days. Well-being rose, d = 0.25. Then everyone returned to work, and past the first week back nothing separated them from where they had stood before the holiday at any significant level, every effect size sitting at or under 0.12. de Bloom and colleagues had found the same shape in 2009 across seven studies: a gain during the break, a fade once the job resumed. Taking a longer holiday did not protect the gain.
Rest is not optional and nothing here argues against it. Six days of it is a pause rather than a stimulus, and a pause leaves the gradient exactly where it was. Where the actual problem is knowing what to do and not doing it, a week of switching off hands you back the same life you left, better rested, with every structural feature intact and waiting.
On rest itself the second camp is right. It is answering a question that most of the people arriving from the first camp were never asking.
The programme that cut diabetes by 58 percent, and what was inside it
Whoever believes that structured doing is the gentle option should start with the Diabetes Prevention Program.
3,234 adults whose risk of developing type 2 diabetes was already high were randomised to placebo, to metformin, or to an intensive lifestyle programme, with the results carried by the New England Journal of Medicine in February 2002. Across a mean of 2.8 years the lifestyle arm reduced diabetes incidence by 58%, confidence interval 48 to 66. Metformin managed 31%. The blinded phase was halted a year early on the advice of the monitoring board, because the answer had already arrived.
The contents of that lifestyle arm are what gets lost whenever the trial is compressed into the phrase "diet and exercise". Each participant was assigned an individual case manager, in effect a coach, with frequent one-to-one contact. There was a structured 16-lesson core curriculum covering diet, activity and behaviour change, taught individually across the first 24 weeks. The published description of the intervention adds supervised activity sessions and a flexible maintenance phase built with restarts for the people who came off it. The stated targets were unremarkable, 7% weight loss and 150 minutes of activity a week, and participants reached them because the whole apparatus had been engineered around doing instead of knowing.
Nobody in that trial was short of information. What they were given was the structure.
The same principle has a much smaller version with its own literature behind it. Gollwitzer and Sheeran gathered 94 independent tests for Advances in Experimental Social Psychology in 2006 and found that specifying when, where and how in advance, inside a single if-then sentence, landed an effect on reaching the goal in the medium-to-large band, d = 0.65, set against holding that same goal without a plan. One written sentence outperforms a great deal of what the first camp sells for four figures.
Keep the number that answers when you work
If one measurement survives the cull, make it the one that responds to effort.
Cardiorespiratory fitness is the strongest candidate available. Mandsager and a Cleveland Clinic group tracked 122,007 adults who had been referred for exercise treadmill testing across the years 1991 to 2014. Their results appeared in JAMA Network Open in 2018. Across 1.1 million person-years, fitness ran inversely to all-cause mortality with no ceiling that the data could find. Those in the top band, at least two standard deviations above the mean for their age and sex, carried an adjusted hazard ratio of 0.20 against the least fit group, which is roughly an 80% lower risk.
Be exact about what that study is. It is a retrospective cohort of people who had a clinical reason to be on a treadmill, not a randomised trial, and reverse causation sits right in the middle of it, since people who are quietly unwell perform worse on a treadmill for reasons that have nothing to do with training. So it is a strong association and it is not proof of cause. It also points the same way as decades of exercise trials, and unlike nearly everything else in the folder, it is a number a person can move within weeks by doing something hard on purpose.
That is the filter worth applying to every measurement currently running. Does this reading change when I do the work? Where the answer is no, or where nobody has ever bothered to check, what you are holding is a subscription rather than a priority.
The three answers side by side
| Approach | Best evidence for it | What that evidence shows |
|---|---|---|
| Measure more: panels, wearables, scans | Ferguson 2022 umbrella review; Krogsbøll 2012 and 2019 | Behaviour moves, physiology largely does not, mortality does not |
| Rest and switch off | Speth 2023; de Bloom 2009 | Well-being rises, then fades inside the first week back |
| Monitoring tied to a goal and a witness | Harkin 2016 | d = 0.40, larger when written down and reported to someone |
| If-then planning | Gollwitzer and Sheeran 2006 | d = 0.65 over holding the same goal alone |
| Structured doing with a coach | Diabetes Prevention Program 2002 | 58% fewer diabetes cases across a mean 2.8 years |
Where this evidence stops
An argument about measurement that goes vague on its own evidence has earned whatever scepticism follows, so here are the limits at the length they deserve.
The Cochrane reviews cover general health checks offered to broad populations. They are not a verdict on investigating a specific symptom, on established screening programmes with their own trial evidence, or on monitoring a condition somebody already has. Reading them as an argument against seeing a doctor would be a misreading, and a dangerous one.
The imaging figures come with confidence intervals wide enough to drive through. Gibson's 3.9% carries an interval running from 0.4% to 27.1%, and Kwee's pooled 16% rests on six of twelve included studies. That last figure is a proportion of the findings that were reported and verified, and it is not a claim that 16% of people scanned receive a false alarm. Both reviews report heterogeneity high enough that the honest summary is a direction rather than a number.
The IDEA result is one trial in young adults with overweight or obesity, inside a specific behavioural programme, with a web interface that covered diet as well as activity. It does not establish that a wearable makes any intervention worse. It establishes that adding one to that intervention, in that population, did not help.
The Diabetes Prevention Program tested a population at high risk of a specific disease and measured a specific endpoint. Its 58% belongs to that trial and does not transfer wholesale to a general adult wanting to feel better. What transfers is the structural finding, which is that the arm with coaching, curriculum, supervision and a restart mechanism beat the arm with a drug, and beat it comfortably.
Effect sizes throughout this piece are group averages. A d of 0.40 or 0.65 describes a distribution shifting, never an individual outcome, and anyone can sit at either tail. And the holiday meta-analyses measured well-being, not physiology, over follow-up windows counted in weeks. They say a break fades. They do not say rest is worthless, and reading them that way would invert their own conclusion.
Where Atlas Cove fits
Assessment has not stopped mattering. It has stopped being a lifestyle. Measure properly, once, at a fixed moment, so that whatever follows gets built from a real position rather than an assumed one. Then shut the folder and go and do the work.
That sequence is what the six days in Sesimbra are built around. Day one is measurement on validated instruments. By the next morning those readings have become a written protocol belonging to that person. The days after are the work itself, done with a coach present. Day six repeats the measurements so the change is something a guest reads rather than something we describe. Then the first month at home stays guided, because the week is the easy half and a protocol that cannot survive an ordinary calendar was never going to alter anything. Our line for this is that we don't diagnose, we build.
Atlas Cove was built in Portugal by someone who came to this industry from outside it, and the reason for one small cohort each month across ten rooms, rather than anything that scales, comes down to something unglamorous. Density of coaching isn't a luxury feature bolted onto an offer. In the Diabetes Prevention Program it was the intervention.
Five years in the first camp, and when it fails the instinct says the data must have been the weak link. That is the trap, and the sector is delighted to sell the upgrade. What sits underneath is not an information shortage: the folder is the proof of that. It is an energy account quietly running an overdraft, plus a body described in exhaustive detail while nothing was ever asked of it, plus a stack of intentions that never once turned into a structure holding a coach, a schedule and a witness. Before spending money at the frontier end of this market, it is worth reading what to fix first and in what order.
So the question was never which panel to run next. It is this instead: when did a measurement last change what you actually did on a given Tuesday? Where the honest answer is that none of them ever did, what would it take to spend six days building rather than watching?
Questions we get asked
Do annual health checks reduce the risk of dying?
On the pooled trial evidence, no. Cochrane found no reduction in overall mortality, cardiovascular mortality or cancer mortality across the trials with usable data, and the 2019 update graded the ischaemic heart disease finding as high certainty. That verdict covers general checks offered to whole populations, and says nothing about investigating an actual symptom.
Do fitness trackers improve health, or only activity?
Mostly the activity. The umbrella review found real behavioural gains, roughly 1,800 steps and 40 minutes of walking a day, alongside effects on cholesterol, glycated haemoglobin and blood pressure that were typically small and often not significant. Systolic blood pressure is the partial exception.
Is a whole-body MRI worth it if I have no symptoms?
On the evidence, a very sensitive scan run over a symptom-free population mostly picks up things that prove to be nothing at all, and every one of them buys a follow-up and a stretch of worry. Serious incidental findings ran to 3.9% of scans in the pooled review, with an interval wide enough that the number should be treated as a range. This is not medical advice and it is a decision to take with a doctor.
If measuring does not change anything, why measure at all?
Because monitoring does work when it is built correctly, at d = 0.40, and the moderators show what correctly means: attached to a goal, physically recorded, and seen by another person. A device silently accruing readings is a different activity with a different evidence base, which is to say none.
Can six days actually change anything?
Six days of rest changes very little, on the holiday meta-analyses. Six days that produce a protocol, a coach and a structure to come home to is a different proposition, and it is the reason the guided month afterwards is part of what gets bought rather than something sold on later. Outcome data from our own cohorts does not exist yet.
Sources
Krogsbøll LT, Jørgensen KJ, Grønhøj Larsen C, Gøtzsche PC. General health checks in adults for reducing morbidity and mortality from disease: Cochrane systematic review and meta-analysis. BMJ. 2012;345:e7191. DOI: 10.1136/bmj.e7191
Krogsbøll LT, Jørgensen KJ, Gøtzsche PC. General health checks in adults for reducing morbidity and mortality from disease. Cochrane Database of Systematic Reviews. 2019;1(1):CD009009. DOI: 10.1002/14651858.CD009009.pub3
Ferguson T, Olds T, Curtis R, et al. Effectiveness of wearable activity trackers to increase physical activity and improve health: a systematic review of systematic reviews and meta-analyses. The Lancet Digital Health. 2022;4(8):e615-e626. DOI: 10.1016/S2589-7500(22)00111-X
Jakicic JM, Davis KK, Rogers RJ, et al. Effect of wearable technology combined with a lifestyle intervention on long-term weight loss: the IDEA randomized clinical trial. JAMA. 2016;316(11):1161-1171. DOI: 10.1001/jama.2016.12858
Gibson LM, Paul L, Chappell FM, et al. Potentially serious incidental findings on brain and body magnetic resonance imaging of apparently asymptomatic adults: systematic review and meta-analysis. BMJ. 2018;363:k4577. DOI: 10.1136/bmj.k4577
Kwee RM, Kwee TC. Whole-body MRI for preventive health screening: a systematic review of the literature. Journal of Magnetic Resonance Imaging. 2019;50(5):1489-1503. DOI: 10.1002/jmri.26736
Webb TL, Sheeran P. Does changing behavioral intentions engender behavior change? A meta-analysis of the experimental evidence. Psychological Bulletin. 2006;132(2):249-268. DOI: 10.1037/0033-2909.132.2.249
Harkin B, Webb TL, Chang BPI, et al. Does monitoring goal progress promote goal attainment? A meta-analysis of the experimental evidence. Psychological Bulletin. 2016;142(2):198-229. DOI: 10.1037/bul0000025
Speth F, Wendsche J, Wegge J. We continue to recover through vacation! Meta-analysis of vacation effects on well-being and its fade-out. European Psychologist. 2023;28(4):274-287. DOI: 10.1027/1016-9040/a000518
de Bloom J, Kompier M, Geurts S, de Weerth C, Taris T, Sonnentag S. Do we recover from vacation? Meta-analysis of vacation effects on health and well-being. Journal of Occupational Health. 2009;51(1):13-25. DOI: 10.1539/joh.K8004
Knowler WC, Barrett-Connor E, Fowler SE, et al; Diabetes Prevention Program Research Group. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. New England Journal of Medicine. 2002;346(6):393-403. DOI: 10.1056/NEJMoa012512
Diabetes Prevention Program (DPP) Research Group. The Diabetes Prevention Program (DPP): description of lifestyle intervention. Diabetes Care. 2002;25(12):2165-2171. DOI: 10.2337/diacare.25.12.2165
Gollwitzer PM, Sheeran P. Implementation intentions and goal achievement: a meta-analysis of effects and processes. Advances in Experimental Social Psychology. 2006;38:69-119. DOI: 10.1016/S0065-2601(06)38002-1
Mandsager K, Harb S, Cremer P, Phelan D, Nissen SE, Jaber W. Association of cardiorespiratory fitness with long-term mortality among adults undergoing exercise treadmill testing. JAMA Network Open. 2018;1(6):e183605. DOI: 10.1001/jamanetworkopen.2018.3605
This is an educational and strategic perspective, not personal medical advice. The views are the author's own and not statements by Atlas Cove Lda.
Lisa Wuerden · Co-Founder
Co-founder, brand and product